This article helps Admins and Community Managers set up and configure Gainsight Community MCP in Claude.
Overview
Gainsight Community MCP is a connector that lets you interact with your community data through AI assistants such as Claude. Model Context Protocol (MCP) is a standardized way for AI assistants to connect to external tools and data sources. Once connected, you can ask questions in natural language and get answers drawn directly from your community.
Setting up the Gainsight Community MCP involves two parts:
- Admin Setup: An admin enables Community MCP for the community from Control. This is a one-time configuration step.
- User Connection: Each user connects from their AI client, such as Claude, using their community credentials. The tools available to each user depend on their community role and permissions.
The scope of tools available depends on your community role. When you connect, the connector authenticates you against your community and displays only the tools that match your role. The tools are organized across three areas:
- Search Tools: Find community content such as posts, topics, and discussions.
- Analytics Tools: Query community performance metrics and dashboard KPIs.
- Admin Tools (Read-only): Review community state such as moderation queues, recognition data, user information, community structure, and idea workflows.
The AI assistant selects the appropriate tools based on the prompts. Since the available tools depend on the community role assigned to each user, reviewing tool access helps determine what types of questions to ask and what data to expect.
- Search to find content
- Analytics to understand trends
- Admin to see what needs your attention.
Prerequisites
Before you begin, ensure the following requirements are met:
- You must have an Admin, Community Manager, or a Moderator role in your Gainsight Customer Community to enable the Community MCP server in Control.
- The community must have at least one authentication method configured (SAML, OIDC, OAuth2, Google SSO, or local username and password).
- Users who will connect must have an active account in the community with a valid role assignment.
- Users must have access to an AI client that supports MCP connectors, such as Claude.
- If your organization restricts outbound network traffic, ensure that your AI client can reach the Gainsight Community MCP server at https://mcp.insided.com/mcp. Refer to your AI client's documentation for the current IP range list.
Limitations
The following limitations apply to the Gainsight Community MCP:
- Admin tools are read-only. You cannot modify community data through the MCP connector.
- After you turn on or turn off the Community MCP toggle, it may take some time for the change to take effect.
- You can connect only one community at a time.
Enable MCP in Customer Community
Admins must turn on the Community MCP Server toggle in Control before users can connect from their AI clients. This is a one-time setup step.
To enable Community MCP:
- Log in to Control.
- Navigate to AI > AI Settings. The AI Settings page appears.
- Scroll down to Features and turn on the Community MCP toggle.
Note: After you turn on the toggle, it may take some time for the change to take effect.
Configure Gainsight Community MCP in Claude
Once Community MCP is enabled in control, admins need to set up the configuration in AI Assistant.
The connector uses dynamic client registration with PKCE, so no OAuth client credentials, authorization URLs, or callback allowlists are required at any point.
The following steps describe the connection flow using Claude as an example.
To configure Gainsight Community MCP in Claude:
- Log in to Claude.
- Navigate to Customize > Connectors. The Connectors page appears.
- Click the Plus icon.
- Select Add Custom Connector. The Add Custom Connector dialog appears.
- In the Name textfield, enter the name to identify the connector. For example Gainsight Community MCP.
-
In the Remote MCP URL textfield, enter the MCP URL
https://mcp.insided.com/mcp
Note: No OAuth client credentials are required.
- Click Add. The MCP connector is now added in Claude.
- Click Connect. The Select Your Community page appears.
-
In the Community URL textfield, enter your valid community URL. For example, if your community URL is
https://community.acme.com, entercommunity.acme.com
- Click Continue. The community Login page appears.
- Enter the following details:
- Username: Enter a valid user name.
- Password: Enter your community password.
- Click LOG IN. The Authorize Application dialog appears.
- Click Approve. The Login is successful.
The permissions displayed depend on your role in the community.
The following table describes the permissions displayed for each role level:
| Role | Permissions Displayed |
| All users | View your profile information, Search community content, View your email address |
| Admins and Community Managers | All user permissions, plus: View community analytics |
| Admins | All user permissions, plus: View community analytics, Manage community settings |
| Moderators | All user permissions, Manage Community settings |
You are redirected back to Claude. The connector now shows as connected with a list of available tools based on your role.
Tools Available in Gainsight Community MCP
The Gainsight Community MCP provides a set of tools organized across three areas:
- Admin
- Search
- Analytics
The tools available to you depend on your community role and the scopes granted during the connection flow.
The following table summarizes the scopes and tools available for each role:
| Role | Scopes Granted | Tools Visible |
| Admins | mcp:search, mcp:analytics, mcp:admin | All search, analytics, and admin tools |
| Community Manager | mcp:search, mcp:analytics, mcp:admin | All search, analytics, and admin tools |
| Moderator | mcp:search, mcp:admin | Search and admin tools |
| Registered user, Super-user | mcp:search | Search tools only |
Admin Tool
The Admin Tool gives community admins a conversational way to review community state. It includes four read-only tools that surface information without modifying any community data.
The following table describes each Admin tool and the types of queries it supports:
| Tool | What You Can Ask |
| Moderation Overview | Surface the moderation queue state, including trashed content and content with moderation labels. |
| Recognition Overview | Read leaderboards (all-time and weekly), available badges, an individual user's leaderboard position, and points history. |
| User Overview | Look up users by ID or field (email, username) and view their current role assignments. |
| Community Structure | Navigate the category tree, view content per category, and check visible topic counts. |
The following scenario demonstrates how the Admin tools work together in a single conversation:
- Prompt: What needs my attention today, and who should I celebrate this week?
Result: The Moderation Overview surfaces content that has been trashed and content with active moderation labels, and the Recognition Overview returns this week's top point earners so you can recognize them publicly.
Search Tool
The Search Tool allows you to query community posts, topics, and indexed content through the MCP connector. The AI assistant retrieves this data and uses its own capabilities to generate insights such as sentiment analysis across discussions, activity dashboards, member engagement breakdowns, and classification of search results by type.
Example prompts:
- What were the most discussed topics this month?
- Run a sentiment analysis on posts from the last 30 days.
- Classify recent search results by type (questions, complaints, feature requests).
- What were the most recent topics in the community?
Analytics Tool
The Analytics Tool allows admins and Community Managers to query community analytics data conversationally through the AI assistant. Previously, reviewing community KPIs required navigating to individual dashboards and manually comparing metrics across time periods.
The following table describes the insight areas available through the Analytics Tool:
| Insight Area | What You Can Learn |
| Community Health | Overall community performance with period-over-period comparison across users, content, traffic, Q&A, and gamification. |
| User Insights | Individual user profiles and activity details. |
| User Trends | Registration trends, leaderboards, role breakdowns, active and churned user lists, and subscription activity. |
| Content Performance | Topic and reply analytics, Q&A metrics, response times, helpfulness ratings, and content gaps such as unanswered questions. |
| Audience Traffic | Pageview trends, visitor segments (contributors, reactors, viewers), and breakdowns by section or content type. |
| Product Ideas | Idea pipeline status, top-voted ideas, and delivered ideas. |
Example prompts:
- How is the community performing this quarter compared to last quarter?
- How many new users joined in the last 30 days?
- Which categories have the most unanswered questions?
- Who are the top contributors this month?
Note: The AI assistant retrieves data that matches the KPIs visible in your community dashboards. The accuracy of the results depends on the quality and completeness of the data in your community.
